Product analytics tracking audit
Tested event tracking across six SaaS products in production to find where analytics events were lost and why.
Code available on request
17root causes found across 65 surfaces tested
Context
Product decisions depended on analytics data that nobody had checked end to end across six products.
What I did
- Tested 65 production surfaces and recorded every tracking call and whether it was delivered.
- Grouped the findings into 17 root-cause issues, each with severity, steps to reproduce, expected and actual behaviour, and separate actions for engineering and QA.
- Split "no tracking" into three separate problems: delivery failures, intermittent firing and missing instrumentation. Each needs a different fix.
- Re-tested every finding before submitting, and corrected my own earlier conclusions where the new evidence required it.
Key decisions
Report by root cause over one row per failing page
Engineers can act on 17 causes. A list of 65 pages is much harder to act on.
Result
Traced most of the lost events to a single delivery path, which gave engineering a clear starting point. Delivered an executive summary alongside the engineering report.
Next steps
Check backend events against the data warehouse, which browser-level testing cannot see.